a

ape/llm.py

@@ -12,6 +12,7 @@ from langchain.prompts.chat import (
 from langchain.prompts import PromptTemplate
 from langchain.chains.llm import LLMChain
 
+#默认使用davinci-003来测试和评估（可控性高），使用ChatGPT生成指令（便宜）
 Cost = {
     'davinci': 0.02,
     'chatgpt': 0.004
@@ -23,7 +24,7 @@ class LLMGPT(object):
 
     def __init__(self, openai_key):
         self.gen_llm = ChatOpenAI(openai_api_key=openai_key, max_tokens=2000, temperature=0.7, verbose=True)
-        self.eval_llm = OpenAI(openai_api_key=openai_key, max_tokens=2000, temperature=0.7, logprobs=1, verbose=True)
+        self.eval_llm = OpenAI(openai_api_key=openai_key, max_tokens=2000, temperature=0.7, logprobs=2, verbose=True)
         self.test_llm = OpenAI(openai_api_key=openai_key, max_tokens=2000, temperature=0.7,  verbose=True)
         self.gen_chain = None
         self.eval_chain = None

app.py

@@ -10,19 +10,19 @@ with gr.Blocks(title="Automatic Prompt Engineer", theme=gr.themes.Glass()) as de
     gr.Markdown("# Automatic Prompt Engineer")
     with gr.Row():
         with gr.Column(scale=2):
-            gr.Markdown("## Configuration")
+            gr.Markdown("## 第一步：输入参数")
             with gr.Row():
                 openai_key = gr.Textbox(type='password', label='输入 API key')
             with gr.Row():
-                n_train = gr.Slider(label="Number of Train", minimum=1, maximum=20, step=1, value=5)
-                n_few_shot = gr.Slider(label="Number of FewShot", minimum=1, maximum=20, step=1, value=5)
+                n_train = gr.Slider(label="训练样本数", minimum=1, maximum=20, step=1, value=5)
+                n_few_shot = gr.Slider(label="每组几个样例", minimum=1, maximum=20, step=1, value=5)
 
             with gr.Row():
-                n_eval = gr.Slider(label="Number of Eval", minimum=1, maximum=30, step=5, value=20)
+                n_eval = gr.Slider(label="评估样本数", minimum=1, maximum=30, step=5, value=20)
 
         with gr.Column(scale=3):
-            gr.Markdown("## Load Data")
-            with gr.Tab("Choose Dataset"):
+            gr.Markdown("## 第二步加载数据")
+            with gr.Tab("选择数据"):
                 with gr.Row():
                     file = gr.File(label='上传txt文件，input\toutput\n', file_types=['txt'])
                 with gr.Row():
@@ -34,14 +34,14 @@ with gr.Blocks(title="Automatic Prompt Engineer", theme=gr.themes.Glass()) as de
                     sample_button = gr.Button('sample Data')
                     sample_flag = gr.Textbox()
 
-            with gr.Tab("Display Sampled Dataset"):
+            with gr.Tab("展示数据"):
                 with gr.Row():
                     train_str = gr.Textbox(max_lines=100, lines=10, label="Data for prompt generation")
                     eval_str = gr.Textbox(max_lines=100, lines=10, label="Data for scoring")
 
     with gr.Row():
         with gr.Column(scale=2):
-            gr.Markdown("## Run APE: 可以替换以下默认指令模板")
+            gr.Markdown("## Run APE: 可以替换以下默认指令")
             gen_prompt = gr.Textbox(max_lines=100, lines=3, interative=True,
                                     placeholder=MyTemplate['gen_user_prompt'],
                                     value='', label="Prompt for generation")
@@ -60,13 +60,13 @@ with gr.Blocks(title="Automatic Prompt Engineer", theme=gr.themes.Glass()) as de
                 eval_button = gr.Button("Eval")
 
         with gr.Column(scale=3):
-            gr.Markdown("## Get Result")
-            with gr.Tab("APE Results"):
+            gr.Markdown("## APE 结果")
+            with gr.Tab("生成指令"):
                 all_prompt = gr.Textbox(label='Generated Prompt')
                 # Display all generated prompt with log probs
                 output_df = gr.DataFrame(type='pandas', headers=['Prompt', 'Likelihood'], wrap=True, interactive=False)
 
-            with gr.Tab("Test Prompt"):
+            with gr.Tab("指令单测"):
                 # Test the output of LLM using prompt
                 with gr.Row():
                     with gr.Column(scale=1):
@@ -76,7 +76,7 @@ with gr.Blocks(title="Automatic Prompt Engineer", theme=gr.themes.Glass()) as de
                     with gr.Column(scale=1):
                         test_output = gr.Textbox(lines=9, value="", label="Model Output")
 
-            with gr.Tab("Eval Prompt"):
+            with gr.Tab("指令评估"):
                 # By Default use the Evaluation Set in APE
                 with gr.Row():
                     with gr.Column(scale=1):